Skip to main content

Today’s column is written by Marc Rossen, SVP investment and activation analytics at Omnicom Media Group.

Clean rooms are emerging to provide the advertising industry a life raft in the era of cookie deprecation.

On the one hand, clean rooms are significantly better than relying on cookie-based ad server data. They enable more stable connections to consumers and their media engagement behavior for each supply side partner. However, the unintended consequence is fragmentation. While it’s possible to dig deep into a single source, it’s much more challenging to look across the media ecosystem to deliver insights, optimization and incrementality.

Taking the “room” metaphor literally, the current system is akin to a collection of stand-alone rooms existing outside the form and function of a house. That’s why the next level must look beyond clean rooms. It’s time to build a clean house using a method called distributed analytics.

Distributed analytics: Tools for building a clean house

Distributed analytics is a technical term that refers to spreading data analysis workloads over multiple clean rooms where media investments are taking place. When applied to clean room data, distributed analytics enables a reunification of holistic media data.

First, audience-based data is gathered just as it would be for traditional planning and activation. Then, audience definitions are carried into each partner’s clean room with media engagement data from a brand’s media campaign. Each clean room’s media data is extracted using the defined audiences to establish commonality. Finally, machine learning is applied to reunite the data into a holistic view of consumer behavior. This enables a variety of planning, optimization and measurement use cases.

The result is brands have a better understanding of their customers. They can better plan, activate, optimize and measure in an ecosystem where first-party identity connections are durable. There’s also richer resulting data from media partners that marketers traditionally could not access. For example, brands can now analyze purchase behaviors from activities like in-store pickup or same-day delivery.

Brands that begin to center their media investments and activation strategy on this stable and richer data today will have a competitive advantage in building better relationships with their consumers.

Developing a distributed analytics capability

Building a distributed analytics capability requires three key elements:

  • Talent: You’ll need practitioners in marketing technology, a scaled team of data scientists and either an external or in-house cloud technology team to manage cloud infrastructure.
  • Tech stack: Every client situation is different, but a centralized, neutral clean room is fundamental. It must be able to connect to internal platforms like CDPs and external platforms via APIs to collect media partner clean room data through data sharing methods.
  • Analytics development: The data and technology is only as good as the applied machine learning using Distributed Analytics methods. This R&D will need planning with a data science team and closely coordinated with media activation and investment functions.

Brands that can harness the power of data clean rooms with applied distributed analytics will be better able to maintain a direct conversation with their consumers across any media vehicle in the coming cookieless ecosystem.